A clinicopathological prediction model for recurrence in patients with early-stage hormone-receptor-positive, HER2-negative breast cancer at King Chulalongkorn Memorial Hospital.

Abstract

e12015 Background: Genetic recurrent score has been increasingly applied in combination with clinicopathology for predicting recurrence and guiding the potential role of adjuvant chemotherapy in patients with early breast cancer. However, lack of accessibility limits its use for clinical practice especially in developing countries. Our study aims to formulate model for predicting risk of recurrence in patients with hormone-receptor (HR)-positive, human epidermal growth factor receptor 2 (HER2)-negative early breast cancer. Methods: Five hundred and twenty nine HR-positive, HER2-negative early breast cancer patients who were diagnosed during 2005 to 2015 in King Chulalongkorn Memorial Hospital were retrospectively reviewed. All potentially clinicopathologic prognostic factors for recurrence were analyzed using multivariable logistic regression. Results: Ninety eight HR-positive, HER2-negative early breast cancers developed distant recurrences after a median follow-up of 65.7 months (IQR, 41.2 - 92.3). There was no difference among recurrent and non-recurrent groups in term of age and menopausal status. However, greater number of patients in non-recurrent group had earlier stage (stage I or II). Most of patients in both groups received hormonal therapy, while more patients in the recurrent group received adjuvant (82.7% vs. 70.8%; P = 0.017) and neoadjuvant chemotherapy (35.1% vs. 11.2 %; P < 0.001). From multiple logistic regression using clinical and pathologic variables, lymphovascular invasion (LVI), percentage of Ki-67, and tumor size were independent predictors of recurrence (OR 3.04; P < 0.001, OR 1.02; P = 0.001, and OR 1.18; P= 0.05, respectively). The prediction model for 10-year recurrent risk appears as “log (odd) = - 3.535 + 0.020 (percentage of Ki-67) + 0.165 (Tumor size in cm) + 1.111 (LVI)”. Probability of recurrence is defined as odd / (1+odd). This model yields a C statistic above 0.7 on validation in independent population. Conclusions: A new clinicopathologic prediction model provides comprehensive recurrent risk estimation for Thai patients with HR-positive, HER2-negative early breast cancer.